IRC channel logs
2026-04-27.log
back to list of logs
<dajole>I want to write a simple service to start my synapse server. However, reading the service documentation I'm sill left utterly confused. E.g. it says that each service has to have a service extension. But why? And how do I know which one? <PotentialUser-76>hello, i am enjoying usin guix, but i am wondering if guix is a welcoming distro ? my grilfriend wants to use guix, and her install can't play music, should i tell her to use fedora instead ? <macabro>PotentialUser-76: I had similar issues but only when I went with a window manager instead of a full desktop environment <PotentialUser-76>we are looking at the fact that we probably didn't set a variable called alsa-service-type <PotentialUser-76>is there a place we can find some kind of default config file so we can compare to ours ? <nckx>sneek: Later tell Guest27 [to the Guest dealing with bcachefs] No, it is not expected that it's mounted read-only or root-only, and there's nothing in Guix that would do that (bug or not AFAIK). This isn't FUSE with its weird semantics. In your case I would first triple-check that my permissions are correct—it's perfectly possible to chmod a file system to allow writes by root only, no bugs required. <nckx>sneek: Later tell Guest27 [to the Guest dealing with bcachefs] Also, probably best to move this to help-guix@. <nckx>sneek: botsnack if you can deliver that without truncating or otherwise screwing it up, my sweet. <nckx>But instead you'll just deliver it to the first unsuspecting GuestNNN regardless of their nickname, won't you. <apteryx>noe: hello! I'll finalize your committer access now; you don't use gnupg subkeys, do you? <efraim>nckx: sorry, I wandered away. I agree about || hurting on the inside. It's for the patches mrustc applies to the rust sources, where we remove the llvm directory and it is referenced in the mrustc patches and we need to edit those patches. <futurile>travel day for me today - see y'all y'all later <noe>apteryx, I’ve updated the key, thanks for the reminder <neox>civodul, #8040 does not seem to be fixed (I tried to build php and it failed) <gabber`>basicnpc: this is an issue of the nonguix channel (and that substitute architecture). you may have more luck over at #nonguix <gabber`>i unfortunately have no insights into these workings. i'd recommend trying out librewolf from the main guix channel/substitue infra <spiderbit>hmm so currently emacs-telegram is broken... I wonder if I can somehow pin a older version that did build succesful, it's not telegram itself but a dependency it depends on tdlib and tdlib depends on php 8.5.3 that is the one that breaks <spiderbit>I mean there is a version of php-8.5.3 for x86_64 that did build... I don't understand why that get's replaced by a version that doesn't build in the stable branch, I assume security? <ieure>spiderbit, Sure, you can also transform emacs-telegram to use a patched version of php. Looks to me like the failure is caused by a test that uses the network, so either patching out that test or disabling them all. <ieure>spiderbit, Any change to the inputs of php will trigger a rebuild, it's not necessarily caused by a change to the php package itself. <spiderbit>so what would be the easist thing to do to get it run? <spiderbit>I mean I thought there is a bug and that maybe 1-2 days later it gets fixed but it seems to be months in this status... so I am not optimistic <spiderbit>ieure ok and can you assist me in that or point me to the "...fucking manual" :) <ieure>spiderbit, ..except have the conditional at the top check to (and (package? pkg) (string=? "php" (package-name pkg))) ... <ieure>spiderbit, And change the list in ensure-keyword-arguments to (list #:tests? #f) <ieure>That will transform any PHP package to one which does not run tests. <ieure>spiderbit, ...but which applies whatever you call the first function. <ieure>spiderbit, Then, assuming you name the second function `with-fixed-php', change the emacs-telegram package to (with-fixed-php emacs-telegram) wherever you have that specified. <m4xxed>Hi, I am defining a package and realized that there need to be some udev rules in place for it to work. Now I wanted to define those udev-rules via `plain-file` or something like that in scheme within the package directory so that it can be installed after the install phase using `(install-file ...)`, however I am failing at it. Does anybody know <m4xxed>of a place in the guix repo where this has been done? I have tried grepping the whole guix repo and so far without success :/ <spiderbit>ahh you did poste the same file just different line numbers. <ieure>Yes, I linked to the specific functions. <spiderbit>ok yes didn't understand first why you postet the same file twice. <ieure>spiderbit, How are you installing packages? `guix package -i', Guix Home, manifest, something else? <spiderbit>no as system package guix system reconfigure <spiderbit>I might change that if I try to use it in other distros... thought about that but for now system wide <ieure>My approach is a good match for what you're doing, then. <ieure>It's definitely not recommended to put user stuff like that in your system profile. <spiderbit>Yes I last understood that there are 3 ways to install / configure stuff... 1. system 2. user but with home and then 3. package -i was surprised to not see with package the stuff listed that is installed by home... but sure kind of makes sense... <spiderbit>that was german word by word kindish translation but that does not really work in english with that words... :D <ieure>There are other ways, too, like guix shell with profiles, direnv, etc. <spiderbit>true what is sometimes problematic is that stuff crashes when it conflicts so my emacs daemon did not start as home service because I upgraded the system emacs (and some dependencies) <ieure>Yes, ideally you should have a thing and all its plugins/libraries/extensions/whatever in the same profile. Putting emacs in the system profile and some Emacs packages in your user profile will have problems like that. <spiderbit>yes and then not forget emacs has also a package manager :D <ieure>I don't use it anymore, only Guix to get Emacs packages. <spiderbit>", ...but which applies whatever you call the first function." you mean what I want to call the function that represents the package? <ieure>spiderbit, Yes what you call the first function, no, it does not represent the package. <spiderbit>what do you use debpaste in emacs is broken, too for some reason <nox>I'm baffled there is still no standard home emacs daemon in Guix and we all have to cobble individually our own shepherd service <kestrelwx>There's a pull request from quite a while ago, right? <spiderbit>what do you guys use to paste in emacs? debpaste seems bugged <spiderbit>I put that in fixed-emacs.scm in /etc/guix/ how do I import / load that into the config or do all the files in that folder get automatically loaded? <nox>I didn't understand everything the PR raises, but it seems to me the problems come from trying to give a separate set of package to the emacs daemon instead of putting them in home profile (which seems more practical to me). <kestrelwx>spiderbit: Yes, you can look in the Guile manual for how modules work. You would invoke your Guix command with `-L$DIR_WITH_MODULE` and in the module where you need this procedure, you import this module and apply this procedure to the package it would be fixing. <kestrelwx>So you'd replace the mention of `emacs-telegram` with `(with-full-emacs emacs-telegram)`. <spiderbit>I think it's the emacs-telegram-server part so I replace both or only that? <nox>i wonder if i should try my hand at a PR. <ieure>spiderbit, Just put it in your config.scm, above the operating-system. <spiderbit>I don't want to use this -L always whenever I reconfigure my system if possible... <kestrelwx>spiderbit: whichever would be in your profile explicitly, this procedure would recursively replace `php` in inputs with a fixed one. You also don't have to store configuration files in `/etc/`. <spiderbit>I know but what is the benefit of not having it there :) <kestrelwx>Oh, if your user doesn't have write permissions. <spiderbit>emacs-telega / emacs-telega-server is the name not telegram <spiderbit>I have some setup in emacs that asks for sudo automatically if I open such files... but sure it's a slight anoyance to write in the passport over and over... <spiderbit>is there a reason it's called emacs-xyz and lisp-xyz? the package? <ieure>spiderbit, Module. I think it just means "et cetera." <spiderbit>hmm do I have to do it for tdlib instead because php is a dependency of tdlib and tdlib I think of telegaserver... <spiderbit>or do I need it for both packages... building took forever and failed again <ieure>spiderbit, You need to apply it to any package which has php in its input graph. You can map that second procedure across every package in your operating-system if you want to replace it globally. <ieure>Yes, the procedure will replace any php in the graph. <spiderbit>I just don't get I put emacs-telega-server in that so it should then not do the checks... <ieure>spiderbit, In this context, yes. <spiderbit>building /gnu/store/5lxfc09ny9pibgxjqvcyanwlplclqj8l-provenance.drv... <spiderbit>cannot build derivation `/gnu/store/y9bg6cs19hcma2sq27v7yfwdpbm3ip49-tdlib-1.8.62.drv': 1 dependencies couldn't be built <spiderbit>cannot build derivation `/gnu/store/vhzsnz18jchrb853yzyq92a2h30lizny-emacs-telega-server-1.0.0.drv': 1 dependencies couldn't be built <spiderbit>that makes not much sense to me I have emacs-telega-server in (with...) so why would it then check <spiderbit>| 'check' phasebuilder for `/gnu/store/wvs5wc1z4p4mv968r46krb360n8hginh-php-8.5.3.drv' failed with exit code 1 <spiderbit>maybe because I have emacs-telega above without the (with...) <spiderbit>it then crashes and therefor says also for -server crashed because It did not come so far <spiderbit>maybe because I had telega not with... it compiled it as input succesful and now uses it for both... <spiderbit>I mean it compiled it succesful for telega-server <kestrelwx>Can I not have computed file in a `home-files-service-type` extension? <spiderbit>ok it run through now and I also testet it, it works again... thanks a lot ieure <kestrelwx>I see ```guix home: error: reference to invalid output 'out' of derivation``` <RyanO>Hello! I’m having issues with the installer, is this the right place to ask a question and link a photo of some logs? <ieure>RyanO, Depending on the problem, we might redirect you to the help-guix mailing list, but this is a good first stop. <RyanO>Ok lovely, I’ll take a min to recreate the issue and post it here soon, Thank you. <ieure>RyanO, What installer are you using? <mcarmo>thought I'd ask here, but has anyone here ever used guix as just the package manager on a distro like void? <RyanO>So I walked through the graphical installer, hit confirm, and all I get is three messages on the screen: <RyanO>Oh I cant link photos, Well it just says some of the services get started. But then it hangs there for over an hour, no change <ieure>RyanO, IRC doesn't have attachments, you can upload them to a photo sharing site and paste the link. Or use a thing like IRCCloud, which has that functionality. <ieure>RyanO, What installer are you using? <gabber`>RyanO: what hardware are you trying to install the Guix System on? <RyanO>I’m using a t16 lenovo thinkpad. I generated the nonguix iso from their main repo (I hope thats ok to say here lol) and the graphical installer looks the same, I just had a wifi connection this time, <RyanO>does that answer the installer question? <gabber`>RyanO: did you create that ISO from a recent guix checkout i.e. did you run `guix pull` some time recently before creating the ISO? <gabber`>can you switch to another console (by hitting Ctlr+Alt+F2 or another F key) and checking with `top` what is happening while the machine seemingly hangs? <RyanO>I did a fresh arch linux install, installed the guix binary on that OS, did a guix pull, cloned the nonguix repo down, added the channel, did another guix pull, and then ran the guix system image command to get the iso. <ieure>RyanO, Alright, so very recent. <RyanO>ok im seeing some info from top. Let me get it uploaded somewhere <gabber`>depending on your installation choices (and the state of substitute availability) it could be that your machine has to build some artifacts locally. but i am vaguely guessing here <Rutherther>gabber: but you would see that on the screen that it is building something <RyanO>I tried just shutting down after an hour to see what happened. It formatted the disk but no OS was installed. <ieure>RyanO, If you switch to another VT, can you ping? <RyanO>ping -c 3 www.google.com gave me a result <ieure>RyanO, Okay, it sounds like the issue yelninei linked is the cause, can you time-machine back to a good commit to build your installer and see if that works? <RyanO>ok this is great info. Thanks so much for finding this. That will take me a while! lol. I’ll need to reinstall arch with the guix binary. So see y’all in maybe 2 or 3 hours! <RyanO>I need another computer that can generate these iso’s for me. <ieure>RyanO, The Guix n00b to "I have a dedicated Cuirass machine" pipeline is real. <luca>Hi, anyone know where kernel configs are defined for linux-libre? I'd like to set CONFIG_SECURITY_LANDLOCK=y and CONFIG_LSM="landlock" which I've been running locally for a while now and feel safe upstreaming <ieure>luca, (gnu packages linux), look for defconfig <luca>That file is giagantic and there's dozens of mentions of defconfig, so I'm having trouble finding exactly where to throw these flags 😅 <ieure>luca, I think there's some stuff about this in the Guix Cookbook. But I have had a 0% success rate making kernel package variants. <rogerfarrell>Any recommendations for running Guix containers for self-hosted services? <ieure>Big part of that is that I haven't had to care about building a kernel since long before defconfig existed, and I don't really understand what it's for or why it exists. <luca>I just have had trouble finding out how to upstream these flags 😅 <ieure>luca, Oh, I believe there are defconfig files in the repo. No idea how to generate or update one, though. Look at a PR updating to a new kernel version? <luca>Great idea! I found #4681 which adds the defconfigs in gnu/packages/aux-files/linux-libre/ <luca>So if I want a new kernel config should I add it in all the configs that support the flag? Or would it be prefered to only add it to latest or only x86_64 <luca>(my flags are supported by linux 5.13 and above) <Rutherther>luca: what exactly do you have problem with? you just call the customize-linux with "#:configs '("CONFIG_SECURITY_LANDLOCK=y" ""CONFIG_LSM=landlock"))" <ieure>luca, I don't know, you probably want to touch base with the kernel-team, I assume they're going to want to know if the configs are changing for the default kernel(s). <luca>Exactly, upstream this to guix <luca>It's been working just fine for me under regular daily usage <luca>If I want to "touch base with kernel-team", how would I go about that? A PR? Email? Codeberg issue? <gabber`>luca: most efficient is probably email, through the `etc/teams.scm` script <gabber`>or opening an issue (and tagging that team) <gabber`>depending on how relevant for others you regard your issue <luca>It's really low-pri low-impact "issue", more of a feature request <luca>oh I guess codeberg issues are for bug reports. Hmmm <luca>All right, codeberg it is. The mailing list doesn't look like is doing too well, probably under a bit of stress <Rutherther>that happens sometimes, but that doesn't mean the mailing lists do not work <luca>tbf codeberg is a bit under the weather as well <gabber`>ah. GNU infrastructure is under constant stress since LLMs started DDoS-ing their public content.. <gabber`>lists.gnu.org just loaded for me, but it took almost a minute <ieure>I know it's tough out there, but I am baffled that FSF/GNU seems to have deployed no countermeasures against the slopcrawling, even though they're absolutely crushing all their infrastructure. <gabber`>i am wondering how we could manage to get in a situation where legislatives seem so incompetent in (at least trying) to regulate digital affairs <nox>What a fantastic time to be in. <gabber`>ieure: i think they have? or at least tried? though the whole Anubis hack is somewhat questionable from the purist "no unkown code execution on my personal hardware" standpoint <luca>uhhh actually landlock might already be enabled in 6.18. I'll switch off from my custom kernel and test again <ieure>gabber`, I don't see anything like that on ex. https://savannah.gnu.org -- which took so long to load that LibreWolf offered to try over HTTP. So whatever they may have deployed, it doesn't seem to be working. <luca>Yeah fsf has been very hostile towards proof-of-work and anubis especially <ieure>Yeah, I mean, it absolutely sucks. But having your stuff completely stop working sucks way more. <ieure>It's all horrible. The email spam wars come to HTTP. <gabber`>i guess their hands are just tied. it's not like they were incompetent or unwilling. it's just DDoS all day, every day (or at least i imagine it to be that way) <ieure>It sounds exactly like they're unwilling? <gabber`>because they refuse to impose this obscure, client-side js code on users? <ieure>> If we made our website use Anubis, we would be pressuring users into running malware. <ieure>"If the rule you followed brought you to this, of what use was the rule?" <gabber`>your point being that sometimes we just *have to* supply users with uninspectable code we have no control over (malware)? <luca>Not to get political, but anubis is not "obscure". It's popular enough that codeberg uses it <gabber`>luca: we're already political (: but that's ok (at least for me) <luca>And IMO the FSF calling anubis "malware" is unnacceptable behaviour towards another open source project <ieure>gabber`, My point being that if the FSF wants Free Software to succeed, their message and output needs to be accessible, but they are choosing not to use the thing which will restore that accessibility. <gabber`>i know, and i somewhat agree with that analysis <ieure>gabber`, While I hate the whole situation and wish it was not so, it's survive or die, and they are choosing die. <gabber`>but the same argument goes for binary blobs in the kernel or non-free drivers and such. it /would/ enable/empower more people to use (mostly) free software, but only at the cost at which we can not guarantee integrity of the system <luca>die is a strong word, they're chosing to fight on hard mode. With techniques other than proof-of-work it makes their sysadmin job harder. And the "other side" has a lot more resources to spam than the fsf has to keep up <gabber`>ieure: i am not sure not being able to grant infrastructure to free software projects is such a bad thing. IMHO GNU/FSF probably should focus on bigger, political tasks instead of being relatively bad at these tasks. i mean: that's why we switched to codeberg <ieure>gabber`, It does. That's why I'm running full-fat Linux with the blobs. I don't *like* it, I'd prefer fully Free hardware support, but I'll compromise here until that happens. <untrusem>btw codeberg is using iocaine now not anubis <gabber`>ieure: sure, same here. but at the same time many of us are able to run linux-libre on our VMs and other infra and are very content that this is our default, no? <luca>huh? I swear I saw an anime girl today <untrusem>luca: i have not seen that anime mascot for over a month <ieure>gabber`, Sure, yes, but I think that's my exact point, is that this is better left as an individual decision rather than a blanket ban. I am aware that there's no effective way to apply this optionality to something like Anubis, I guess you could have "*.anubis.gnu.org" protected by it and let users choose which to visit. <nox>I agree that anubis is obscure, client-side js sent over. It's a big, big leap to call it malware... <nox>but we do need a better solution. <ieure>nox, Life in prison for any LLM company CEO! <gabber`>ieure: as much as i agree with you, GNU+FSF would not be GNU+FSF if they were so easy to take the pragmatic options when there is a painful but philosophically correct path <nox>ieure, so many expressions of agreement weighing through my mind, and none of them can be publicly expressed <cdegroot>I'm soo sad that everybody is using this pretty nifty /distributed/ version control system and we're all day long "codeberg is down!", "sourcehut is down!", etc. <mightysands>hey, trying to verify the sig for the guix iso file but I keep getting "gpg: Can't check signature: public key not found" <mightysands>I literally copy pasted and ran the import code above character for character <mightysands>but if it's not the RIGHT one, then...what am I supposed to do next here exactly ? <gabber`>cdegroot: this is mostly about project management (issues, pull request) and not so much about distributing the code... <nox>cdegroot: code is decentralized but collaboration isn't yet, sadly <gabber`>mightysands: is pulling the keyring branch and trying again not doing the trick? <mightysands>I'm sorry to say the term "keyring branch" is new to me <cdegroot>nox: I have a prototype somewhere for collab-over-syncthing. I wonder how well that would scale (it even does git-over-syncthing, but that bit I still need to flesh out more :-)) <cdegroot>But frankly, code is much less decentralized than it could be. <gabber`>mightysands: sorry, did not really read your message. which possibly relevant keys show up when you invoke `gpg --list-keys` <nox>cdegroot, syncthing doesn't scale very well in my experience- the best attempt in the "p2p forge" space is radicle right now I think <mightysands>nckx: Ah, yes this is that same error. Thanks for the link ! <gabber`>nckx: i bumped the issue hoping it gets merged soon (: <nckx>YW, sorry that it happened in the first place. Worst place to have a typo in a release 🤷 <cdegroot>nox never heard of Radicle, thanks for the pointer. <nckx>Probably a silly question but I don't know for what to search: is there a way to mark text (here: the Guix ISO signature verification commands and response) as not subject to translation? In such a way that changes don't have to go through the Weblate pipeline at all, but are immediately visible after ‘make doc/guix.info’. <nckx>The answer must be ‘yes’, I just don't know the right question. <gabber`>i'd email the translation team to get a definitive answer (: <nckx>You would, wouldn't you. That's because you're clever. <Air4x>Hi, regardind the broken test for php 8.5.5 I went and checked the github issues and found this "https://github.com/php/php-src/issues/21672". Should the failing tests be removed until the thing is solved upstream? I'm asking mostly because of curiosity. <ieure>Oh, a time bomb. That should be patched in Guix regardless of upstream. <gabber`>Air4x: if you can confirm the resulting package works as intended that should be fine <Air4x>gabber`: I don't really know how. I have php installed in my system because is a dependency of emacs-telega-server. I don't really know to test it. I did try the fix from "https://codeberg.org/guix/guix/pulls/8186" and 'php --version' works, but anything more is outside of my knowledge. <Air4x>Im searching how to use openssl in php, give me an hour <gabber`>Air4x: i guess you could try to fetch data from an HTTPS resource in a php script, maybe? <Air4x>gabber`: I was thinking about something similar, for now I found an example that encrypts and decrypts a simple string. <dcunit3d>someone in the systemcrafters discourse asked about specifying a custom channel with an SSH uri for git, like "git@foo.bar:my/repo". is that possible? i know `guix copy` refers to the user's SSH configuration, but `guix pull` would likely differ, right? <nox>> someone in the systemcrafters discourse asked about specifying a custom channel with an SSH uri for git, like "git@foo.bar:my/repo". is that possible? i know `guix copy` refers to the user's SSH configuration, but `guix pull` would likely differ, right? <nox>You can specify a custom channel with anything Git will accomodate, I've used it with file:/// uris <dcunit3d>i'm just curious, but i don't tests for that in ./guix/tests/channels.scm <nox>but you'll probably need to pass your key somehow, I'm not sure how <dcunit3d>yes, but does guix-pull establish a connection (using a guile port?) and then hand that off to the daemon? <dcunit3d>i think that's what `guix copy` is doing, but `guix pull` ends up building. <nox>I'm not certain. You can look in ./guix/scripts/pull.scm to see how it's implemented. My uninformed intuition would be that the uri is directly passed to the git library. <dcunit3d>hmm idk. the docs for GUIX_DAEMON_SOCKET mention a need for the remote end to have guile-ssh, but i imagine that `guix pull` uses libgit2 or something <nckx>Air4x: That PHP GitHub bug report should™ not be relevant to Guix? It happens to list one of our failing tests but we already 'regenerate-ssl-certs at build time… <Air4x>nckx: The thing is that the regenerate-ssl-certs seems to be not enough. https://codeberg.org/guix/guix/pulls/8186 fix the problem by switching to openssl-3.0, which is EOL before the end of this year. I tried to switch to openssl-3.5 but the problem remains. <Air4x>Removing the tests is the fastest way to make php build, but I do recognize that is not the best way to solve the problem <dcunit3d>i tried using git@host:fdsa/asdf URI and seems to connect fine... not sure what their problem was. <mightysands>Hey, just tried the USB install of guix and as soon as I try booting into the usb, it starts throwing ima: error communicating with the TPM chip errors then throwing a kernel panic <mightysands>The TPM is disabled. It's inactive. Why is guix refusing to boot just because I don't have TPM active ? <dajole>Is there a better way of searching for Guile libraries than something like `guix search guile yaml` (so in this case to find a module to programmatically generate yaml)? <nckx>Air4x: I just mean that the GitHub bug report is only about expired test certs, and unless openssl has changed how time works since 3.0 (I wouldn't put it past them!), that is not why our sni test fails. So yes, regenerating the certs isn't *enough* to make the test pass, but it doesn't make that upstream bug relevant. <nckx>ACTION deployed an untested PHP to production yesterday, yolo. <nckx>Air4x: Also correction welcome if I'm wrong about that. <cdegroot>mightysands: that's more likely a kernel issue rather than a Guix-specific issue. Does the libre kernel even have tpm support? I have no clue :-). But smells like "it is disabled but not in a way that Linux expects it" to me. <nckx>Reading about unexplained errors ‘in eval()'d code on line 15’ just warms the cockles with joy. <mightysands>cdegroot: That's depressing. I'm just wondering why TPM being disabled would stop me from installing guix like this, because if it's stopping me, it must have stopped countless other users if it's a kernel issue. I mean realistically, how many variables can really be present here ? <mightysands>It's either active or inactive. There is no third option in the BIOS, and I imagine some users have reasons to not activate it. <ieure>mightysands, Cursory searches indicate that the error is not fatal and can be ignored. But that clearing the TPM state in your BIOS settings may make it go away. <mightysands>I'm looking for ways to ignore it, but can't find any. All results on DuckDuckGo seem to be reddit and StackXchange users saying "just enable it then" <ieure>mightysands, I believe it is harmless to enable, do you have a specific reason to disable it? <mightysands>ieure: I'll take a look into clearing its status and rebooting, though I don't see how or why that would avert a kernel panic <ieure>mightysands, An error message is not a kernel panic. <ieure>Is your kernel panicing? Or just printing the error you mentioned? <mightysands>ieure: it's because there's writing on my screen with the words "kernel panic" afterwards <ieure>mightysands, They are not necessarily related, the panic could be caused by something happening after the TPM errors are printed. <nckx>But is it related to the TPM errors in any way? <ieure>I'm not saying they're *not* related, just not that they're *not not* related. <mightysands>that's true. In thinking on it further, the two issues seem unrelated, or at least not necessarily related <nckx>Yeah. There are a lot of random errors in the average dmesg, and they regularly happen just before something totally unrelated fails. <cdegroot>just a panic without register dumps and stack traces and whatnot is rare, though. <ieure>mightysands, Do you have any details in the panic? Like cdegroot is saying, there's usually debugging info, like a stack etc. <mightysands>Specifically it says: GC Warning: Could not open /proc/stat, then it gives a Welcome to early boot GUILE message, and a few lines down it says it's waiting for some partitions to appear before a new sudden wall of text pukes out and suddenly I get Kernel Offset: disabled and ---[ end Kernel panic -- not syncing: Attempted to kill init! exit code=0x0000000 <ieure>Stuff's definitely horked, but if you got to early boot Guile, the kernel is running. <Air4x>nckx: honestly I don't know, I'm just trying to get php to build so that I can finally reconfigure my home environment. <ieure>At least I think it has to be running at that point. <nckx>ACTION .oO When ‘make’ing Guix, ‘guile \ /home/nckx/guix/scripts/guix {environment,home} --help’ hog the CPU for *ages* (*minutes*!), which makes me wonder if they are interpreting code that would be compiled if they'd just wait… <nckx>Air4x: Well, does Home support passing packages by value like (package (inherit php) (arguments (substitute-keyword-arguments arguments ((#:tests? _ #f) #f)))) ? <nckx>That's the new & improved PHP version all my machines are currently running… <mightysands>ieure: I'm really not sure what the problem is. I've never even used this laptop. I just dd'd the guix system iso onto the usb and plugged it in, and selected it as the boot device <ieure>nckx, Yes, this is the normal way Home works. <nckx>ACTION kills HELP2MAN doc/guix-{environment,shell}.1 after many minutes of it heating the room… that's not… right. This is how unrelated debugging in Guix usually ends. <cdegroot>One option is that the laptop requires something that's not in the libre kernel. I mean, the Guix ISO should work just fine, I never had issues there. Missing drivers don't normally cause panics so I do wonder what else. Maybe it's simply a memory corruption issue? <cdegroot>(unexplainable panic -> run memtest86 and see tomorrow :-)) <mightysands>That sounds like an idea worth trying cdegroot. I'll run memtest tomoz and see what it pulls up <ieure>Agree that memtest seems like the right course. 